• Abstract and Figures
• Public Full-text

ON REPRESENTATIONS ATTACHED TO SEMISTABLE VECTOR BUNDLES ON MUMFORD CURVES GABRIEL HERZ Contents Introduction 1 1. Preliminaries 4 1.1. Notations and conventions 4 1.2. Reduction and formal rigid spaces 5 1.3. GAGA-results 8 1.4. Fundamental groups of analytic manifolds 13 1.5. Mumford curves 19 2. Representations attached to vector bundles 23 2.1. A review of the constructions 23 2.2. Comparison of the constructions 31 3. Illustrations 38 3.1. Fundamental groups of a Tate curve 38 3.2. Vector bundles on a Tate curve 39 3.3. Vector bundles on a Mumford curve of genus 2 40 References 41 Introduction A classical result by Andr´eWeil, proved in 1938, asserts that a holo- morphic vector bundle on a Riemann surface is given by a representation of the fundamental group if and only if each indecomposable component is of degree zero (cf. [Wei38]). Furthermore it is known that unitary repre- sentations of the fundamental group are in one-to-one correspondance to polystable vector bundles of degree 0 (cf. [NS65]). arXiv:math/0507407v2 [math.NT] 30 Jan 2006 In 1983 Gerd Faltings introduced the notion of φ-bounded representations and proved a corresponding result in p-adic analysis for vector bundles on a Mumford curve over a discrete non-Archimedean field. He proved an equi- valence of categories between semistable vector bundles of degree zero on a Mumford curve and φ-bounded representations of its Schottky group. In his proof he used the theory of formal schemes and was therefore limited to discrete fields. In 1986 Marius van der Put and Marc Reversat generalised Faltings’ result to arbitrary non-Archimedean fields by using methods from Date: November 29, 2017. This is a slightly modified version of my PhD-thesis at the University of M¨unster under supervision of Christopher Deninger. 1 2 GABRIEL HERZ rigid geometry. Unfortunately the functor they constructed does not com- mute with duals or tensor products. In the case of line bundles it is easy to see that this is inherently caused by the definiton of φ-boundedness. In the sequel we will call their representations PR-representations. Let X be a projective, smooth and geometrically connected algebraic curve over a finite field extension of Qp. In 2004 Annette Werner and Christopher Deninger constructed an ´etale parallel transport for vector bundles with po- tentially strongly semistable reduction on XCp (cf. [DWc]). Restricted to the algebraic fundamental group this is a functor that associates a continuous Cp-vector space representation (in the following called DW-representation) of the algebraic fundamental group of XCp to every vector bundle of this class. This functor is Cp-linear, exact and commutes with duals, tensor products, internal homs and exterior powers. On can pose the question, whether the category of vector bundles with potentially strongly semistable reduction is equal to the category of semistable vector bundles of degree zero. In this thesis we will only be concerned with a special case of the Deninger-Werner construction. Recently Gerd Faltings had announced a p-adic version of non-abelian Hodge- theory. In [Fal03] he defines a category of certain generalised representations and proves an equivalence of categories between them and vector bundles on XCp endowed with a p-adic Higgs field. In his proof he uses the theory alg of almost ´etale extensions. The representations of π1 (XQp ,x) form a full subcategory of Faltings’ generalised representations and Faltings suggests that under the equivalence of categories semistable vector bundles of degree alg zero come from representations of π1 (XQp ,x). In this thesis we will not rely on the greater generality of Faltings approach. Let X be a Mumford curve over a finite field extension of Qp. In this thesis we compare the DW-representations attached to a class of semistable vector bundles of degree zero on XCp to the PR-representations defined for this class of vector bundles. Performing this, obvious problems occur: (1) it is not known whether the DW-representation does exist for all semistable vector bundles of degree zero. (2) Different groups are represented. (3) The functors of Deninger–Werner and van der Put–Reversat have different properties. For example the DW-functor commutes with duals and tensor products, the PR-functor in general does not. The solution of all of these problems is (1) to consider only the subcategory of semistable vector bundles of degree 0 that have a vector bundle model on the extension of the minimal regular model of X to the ring of integers o of Cp, (2) to introduce the notion of finite topological coverings and the finite topological fundamental group of X (3) and at last to prove that the DW-representation factorises through this fundamental group. Having done this, we can prove that the DW-representation and the pro- finitely completed PR-representation attached to vector bundles in the pre- viously mentioned class are isomorphic. The PR-representations attached VECTORBUNDLESONMUMFORDCURVES 3 to vector bundles in this class are isomorphic to representations that have image in GLrk(o) if rk is the rank of the vector bundle considered. At least for line bundles the previously mentioned class is the best possi- ble on which both representations agree, since for line bundles whose PR- representation is not represented by numbers of norm equal to one, the PR-representation does not commute with tensor products and duals, but the DW-functor does. Introducing the fundamental group of finite topological coverings suggests itself, since we have to compare representations of the topological fundamen- tal group (which is the Schottky group) with representations of the finite ´etale fundamental group. The coverings that are topological and finite ´etale are exactly the finite topological coverings. We prove our result by extensive use of GAGA theorems between rigid, for- mal and algebraic geometry. Because we have some GAGA results only in the case of discrete valuation we reduce the case that the vector bundle is only defined after extension to Cp to the case that it is already defined over a discrete valuation field by an argument from non-abelian cohomology. As an illustration we will have a closer look at Mumford curves of genus 1 and 2 and at vector bundles on them. This thesis is organised as follows. In the first section we remind the reader of some notions of rigid and analytic spaces, which are important for us. We cite some GAGA results and prove slight extensions of them. Especially quotients of schemes by finite groups are considered more closely. We state the basic notions of Galois theory, introduce the finite topological fundamental group and prove that it satisfies the six axioms of Grothendieck for a Galois theory. At the end of the first section we introduce Mumford curves and discuss their stable and minimal regular model. In the second section we describe the constructions of van der Put and Reversat, Faltings and Deninger–Werner. We characterise the semistable vector bundles of degree zero that have a vector bundle model on the mini- mal reguar model of the Mumford curve by their PR-representation and we compare the construction of Faltings with the one of van der Put–Reversat. In the next subsection we compare the DW-representation with the PR- representation. We deduce the case that the vector bundle is only defined after base change to Cp from the case that it is already defined over a dis- crete valuation field and prove this case first. In the last section we give some illustrations. We study the various Galois groups of a Tate curve and investigate vector bundles on Tate curves and on Mumford curves of genus 2 in more detail. I would like to thank my supervisior Prof. Christopher Deninger and Prof. Annette Werner for introducing my to this interesting topic. The final version benefited from discussions with Jan Kohlhaase, Sylvain Maugeais, Roland Olbricht, Matthias Strauch and Stefan Wiech. I am grateful that Sylvain Maugeais and Stefan Wiech read a preliminary version. This thesis was partially supported by the Deutsche Forschungsgemeinschaft at the SFB 478 Geometrische Strukturen in der Mathematik. 4 GABRIEL HERZ 1. Preliminaries 1.1. Notations and conventions. We use the theories of schemes, formal schemes, rigid spaces and Berkovich spaces and assume that the reader is familiar with these theories. In particular for the theory of rigid spaces the reader might consult [BGR84], for Berkovich spaces [Ber90] and [Ber93], for formal schemes the reader might refer to [EGAn]. A field endowed with a non-Archimedean valuation that is complete with respect to this valuation is called non-Archimedean field. A discrete non- Archimedean field is a non-Archimedean field whose valuation is discrete. We will assume all non-Archimedean fields to have non-trivial valuation and to be of characteristic zero. The following notations will often be used. The term M(A) denotes the Berkovich spectrum of a commutative Banach ring A with unit, that is it denotes the set of all bounded multiplicative semi-norms on A provided with the weakest topology with respect to which all real-valued functions of the form | · | 7→ |f| (f ∈ A) are continuous. Let K be a non-Archimedean field. The terms K-analytic space and strictly K-analytic space denote the Berkovich analytic spaces that are defined in [Ber93, page 22]. A K-analytic space is called good if every point has an affinoid neighbourhood. The term analytic space always means Berkovich analytic space. A Hausdorff topological space is called paracompact if every open covering has a locally finite refinment. A rigid space is called quasiseperated if the intersection of two open affinoid domains is the finite union of open affinoid domains.

Load more

  42

Copy Link